Problem with Prime95 ECM for M(665280)
 

Prime95 keeps reporting an extra factor of 3, 7, and 11 for the composite exponent 665280. I'm pretty sure M(665280) has only 4 factors of 3, and two factors of 7 and 11. My lowM.txt file has these and many other factors. I'm trying ECM with B1=200 and B2=500, and only 1 curve until I sort this out. One or two of the numbers shows up after the line that starts curve 1, the other factors show up after stage 1 is complete. Which factor shows up where appears random, but all three factors always show up. I'm setting the values through the windows GUI. Using Version 23.6. Using this [url=http://home.ix.netcom.com/~wblipp/lowM.txt]lowM.txt[/url]. I'm out of ideas to try - does anyone see what's wrong?

A new symptom! I deleted all factors above 80 digits and tried again. This time the report was that M(665280) had a factor of 241. One factor of 241 is listed in lowM.txt, and I'm pretty sure there isn't a second factor of 241 in M(665280).

I thought a computer reboot might help. It didn't - I get the same symptons of Prime95 reporting new factors that are already in lowM.txt, and always reporting the same specific factors. Can anyone else duplicate this on their system?

Are the entries in lowm.txt sorted? Prime95 requires this.

[quote="Prime95"]Are the entries in lowm.txt sorted? Prime95 requires this.[/quote]

This was one of my early ideas. Sorting the lowM.txt file didn't change my symptoms. The lowM.txt files I've used most recently are [url=http://home.ix.netcom.com/~wblipp/lowM.txt]this one[/url] and the same file with all factors greater than 80 digits removed. As reported in this thread, the first case reports excess factors of 3, 7, and 11 while the latter case reports an excess factor of 241.

There are only 4 factors of 3. Remove one of these lines and all should be ok.

It seems to be a small bug that prime95 does not raise the "factor in lowm.txt does not divide 2^N-1" message. If you had m(665280) C: 243 it would have caught it.

[quote="Prime95"]There are only 4 factors of 3. Remove one of these lines and all should be ok.[/quote]

:question: I only see four factors of 3 - the first four lines of the lowM.txt file. Am I missing something - perhaps a factor of 3 in one of the supposed primes that follows? (I had trouble with this earlier, but I thought I had fixed it.)

My bad. There are a couple of large factors that are listed twice.

[quote="Prime95"]There are a couple of large factors that are listed twice.[/quote]

:question: I don't see this, either.

I checked all the factors to see if there were any additional factors of 3, 7, 11, or 241 that slipped through. I didn't find any.

M( 665280 )C: 210925997723521
M( 665280 )C: 210925997723521

and

M( 665280 )C: 166242935471754241
M( 665280 )C: 166242935471754241

THANKS! I looked right at those and couldn't see them. That fixed all my anomolous behavior.

Previously duplicated factors had resulted in the message "Factor does not divide N!" It was easier to find the cause then. It would have been even easier yet if the message had told me WHAT factor didn't divide N - I had to resort to binary search in the factor file.

Changing these two things - identifying erronous factors explicitly and uniquely - are on my wish list for some future version, although I understand and agree there are more important things on the "todo" list.